Campaign tracking

Choose the right campaign tracking system and get the most from it

I haven’t met a company that doesn’t struggle with measuring the effectiveness of its marketing campaigns. Growing channel diversity and complex customer behaviour has made campaign analysis for marketing a lot messier.

This article aims to help marketers and their Agencies in three ways:

  • To review the different campaign tracking techniques that are available and how they can be combined and to provide an overview of the strengths and weaknesses of each.
  • To help you decide what level of sophistication is right for your client or company.
  • To help you anticipate some of the issues you will have to address if you want your marketing campaign analysis to be efficient, accurate and actionable

An overview of the key campaign tracking techniques


A review of campaign tracking and measurement solutions

The most commonly used campaign tracking solutions

1. Source Codes

This technique is becomming less popular among heavy direct marketing advertisers although a few do report excellent results. BUPA for example routes calls through a reception centre whose job is accurate source code capture and onward call routing.

However, unlike BUPA, most companies cannot afford nor will their customers tollerate an extra step in the customer journey. Instead companies have spent money training, incentivising and monitoring staff regarding source code accuracy. Unfortunately source codes are simply less important to the customer and the agent than quickly solving the customer’s need.

High levels of miss reporting and 40 – 60% unallocated (‘other’) leads are the result; often even worse when captured by a retail network or input by visitors on a web form.

So for most this campaign tracking technique is unattractive, however it does have two advantages:

  • Firstly source codes are generally input directly into the quote and sales engine allowing response, sales and ROI to be reported in a single view.
  • Secondly it allows a single, highly memorable telephone number to be used on all marketing literature (remember Guardian Direct’s owl logo and 0800 28 28 20 response number?).

To make the most of a Source Code system here are some of the controls you need to put in place:

  • Create a central source code database with a numbering and description convention that enables simple like-for-like comparisons
  • Minimise the granularity or your measurement system to avoid it becoming unwieldy (for example don’t use codes to track performance by list, match-back can do the job far more accurately)
  • Do not expect Agents to select codes from long drop-down lists; use look-up technology to suggest a decreasing list of options as they type in the code
  • Automatically remove old codes and have them default to dump codes
  • Minimise the number of dump codes available
  • Monitor data capture accuracy by Agent
  • Invest in training and incentivisation for accurate data capture
  • Feedback campaign results and changes to the front-line so they understand the value of the MI

2. UTN (Unique Telephone Numbers)

UTNs used on each campaign test cell remove the need for customers to look up codes or Agents to ask for them. This simple step eliminates much of the error found in source coding and allows staff to focus on the customers’ needs. As a result this has become one of the most common techniques used by marketing for campaign analysis in recent years. However it also presents significant challenges.

Most importantly counting the number of calls a test generates does not tell you the number of sales nor the sales value. Call quality varies significantly by medium so applying a flat conversion rate will fatally skew results.

This problem can be reduced; multiple calls can be de-duplicated using caller IDs and minimum call duration can be used as a proxy for call quality or sales.

However the ROI of a test can only be understood if the UTN is appended to the sales record. This can be done automatically (with support from your IT department) or manually by Agents inputting the UTN presented on their screen or whispered into their headset.

The second issue is that people will use the first number that comes to hand, or the number they believe will be answered most quickly, and then use the IVR to transfer to the right area. Either IVR behaviour needs to be well understood to allow proxy sales to be weighted or processes need to be established to append final call destination to the call record.

To get the best from a UTN solution consider the following steps:

  • Develop a central UTN database with naming hierarchy to allow simple like-for-like comparisons
  • Apply date ranges for each UTN campaign with a cut-off date for each to allow UTNs to be recycled efficiently
  • Ensure you have a large pool of UTNs (several hundred for a large DM advertiser)
  • Check to make sure the right UTN has been used before each campaign goes live
  • Create a database to ensure repeat callers are de-duplicated
  • Exclude short calls, and calls routed to non sales area from proxy sales
  • Append UTN and data range to the quote and sales database to generate actual sales and ROI

3. Cookies/PURLs

Online reporting is in many ways far easier than offline reporting as tags are automatically passed from one system to the next, however connecting online and offline activity is notoriously difficult.

Offline response to online activity is often desirable and partial tracking can be created through unique landing pages with UTNs or the use of generic web dump codes by your call centre and retail network.

Where online response to offline activity is required the options include:

  • PURLs (Personalised URLs) for database driven activity. This has good take up where genuinely personalised landing pages are provided and is most useful for high value products as it also allows rapid follow-up to browsing, even when no log-on or form has been completed.
  • Campaign specific URLs or micro sites; these are only effective where offers must be accessed via the destination URL but beware, promoting multiple URLs can have unintended consequences for SEO effectiveness
  • Web analytics solutions using persistent cookies to append browsing history to customer data (if visitors are asked to log on or complete a form)
  • Econometrics for large, intermittent activity

Also you should beware the use of multiple measurement systems for email, banners and PPC which can lead to a highly fragmented view of online response. Most good analytics systems will allow tagging of all online activity and response to be collated into a single application.

4. Customer URN (match-back)

Many use this as their primary measurement technique particularly where response comes via a retail network. However it is a time consuming and slow process that only works for database driven campaigns. Where broadcast and online activity work alongside database marketing match-back over-estimates DM (in the region of 25% – 40% over) to the detriment of broadcast results.

It also means different measurement systems are needed for other media making like-for-like comparison impossible.

Best practice is to measure incremental uplift – sales from a group who received a campaign minus the sales from an identical group who did not.

Another problem to be aware of is that this approach normally only measures sales. It is often valuable to understand response and conversion. Low converting campaigns can incur high handling and opportunity costs and this metric gives insight into problems (or opportunities) in the customer journey.

5. UTN + URN (match-back)

By combining a response and a sales measure companies can get a much fuller picture of the impact of a wide range of activity. This obviously works best where the UTN is appended to the quote and sales engine (see 2 above).

The key draw-backs to this approach are a limited understanding of online and brand activity and that a company must balance two separate versions of the truth and use each one appropriately. I generally recommend ROI by campaign is reported to the business whilst marketing also use conversion rates to understand the attractiveness of different propositions and the effectiveness of the customer journey.

6. UTN + Econometrics

Econometrics will not be appropriate for all companies due to the cost of the modelling and the scale of marketing activity required to generate sufficient data. Most companies embarking down this route will already have other measures in place, however frequently econometrics will be used only to analyse the impact of broadcast activity, ignoring digital and database activity.

Indeed econometrics does struggle to produce meaningful insight into either activity which is either ‘always on’ or is too small to create a sizeable impact. However with care and planning it can provide insight to large parts of the marketing mix, on and offline.

Its value is that is can provide a common currency for businesses to evaluate the full contribution of each element of the mix and to help understand how media work in concert assisting with channel planning. Its weaknesses are that it does not provide a sufficiently granular view to optimise the performance of each channel and that results can be open to misinterpretation.

A mix of UTN and econometrics can provide a common, independent view of the value of each medium, the additive affect of multimedia strategies plus a detailed measure of the response each campaign and test cell produces.

7. UTN, URN (Match-back) and econometrics

This is perhaps the most complete view of marketing performance. It provides a common overview of performance of the media mix, a detailed view of response by test cell and a hard measure of sales and ROI per test cell.

However the high costs and complexity of three versions of the truth mean that most companies would struggle to get value for money from this rich mix of data.

In summary

There is no perfect solution, the right solution for your company or Client will come from understanding the capability of each system and matching it to the needs, resources and business model of the company. Just as important it the realisation that any system can be implemented or managed well or poorly. Understanding the challenges, setting realistic expectations and putting sufficient resource in place to properly manage the chosen campaign tracking system are all necessary to get the most from it. Too many companies unnecessarily switch to a new solution because that which they have is not managed properly or waste money on a new implementation because they have not properly anticipated the steps they need to take.

If you would like more information, or a free impartial view of your marketing analytics challenges then contact me